Changed to basic crystal reference.
I show my RX at ~27.005MHz of the crystal frequency with Park mode 3 at the end of one TX/beginning of another. And two using Park mode 0, one at the start after idle, one between two transmissions.
"Start of TX from Parkmode3" shows a small section of the cyclic movement on 2m.
The initial jump at the start of the transmission is quite fast, probably what Henning sees on 2m.
Personally I do not think the initial rapid frequency shift is of any real consequence for WSPR.
After another hour or so 1032 13 -0.5 144.490529 2 G4ZFQ IO90 0 a drift of 2Hz, not yet as good as Henning but not too bad.
And 1042 13 -0.8 144.490540 0 G4ZFQ IO90 0 Zero drift.
But after that it began to drift again, temperature? Not really, more a slow cyclic drift. Removed Synth module LM317, fitted low drop 3V3 regulator AMS1117, an Ebay module. No difference.
I am thinking this is not entirely a heat-related problem, switching inside the Si5351 as the frequency is changed alters the capacitive loading on the 27MHz crystal? The cyclic drift could be associated with the pattern of WSPR tones.
Now in another thread Hans reminds me that there are no loading capacitors on the crystal (apart from those inside the Si5351 which are lower than specified) Is the crystal therefore more sensitive to loading?
So, in goes my Adafruit module with a 25MHz SM crystal.
This is looking promising. Even though 25MHz means the WSPR frequency shift is compromised it is decoding reliably on 2m with drift 2 or 3Hz. Any fluctuation due to WSPR frequency changes was not obvious.
Now back to the basic 27MHz crystal Synth, put in two 15pF loading capacitors, hoping that would overwhelm the cause of the "modulation" of the 27MHz. It started almost dead on 27MHz, drifting down, settling about 60Hz down. But this was far worse, the 27MHz was not stable even when the U3S was idle.
That is about as far as I can go the Adafruit shows a simple crystal can work on 2m but my Synth board does not.
It is better to use a TCXO and, presumably, the QRP Labs OCXO.
(This is a U3S run with one BS170 and ~5V on the drain so not much heat generated. I have it in a custom box with the display on top, the synth module therefore at the bottom.)
Maybe Park mode needs to be considered carefully? Mode 0 gives an output, my observations see less than 30dB below the WSPR signal level at 144MHz. A setting of 150MHz works OK but if my figure is correct a signal will go out.
> You wonder about the source of 50 Hz sidebands. May I posit a cause?....
Thanks for your thoughts. It is OT if my observations are of locally created spurii but I thought I should mention it so others may check.
I still observe spurii on HF with a different RX system all 50Hz apart, worst -44dB on 10m.
2m not noticed this time, in fact seems fairly clean. Strange, possibly just my test method.
73 Alan G4ZFQ